IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Converting FLAC to MP3 - strange cutoff frequency only when using FB2K
ChesterB
post Feb 10 2013, 03:53
Post #1





Group: Members
Posts: 35
Joined: 2-February 07
Member No.: 40241



Hi, guys, ran into a weird problem and decided to share it with you, hope you can explain this to me. I was converting several lossless rips of albums in FLAC to MP3 (@ 320 CBR) and when checking the spectrogram of one of the resulting files I was quite shocked at what I saw. The song I'll use as an example here is called "Mutant", it's the third song from the album "HyBreed" by the Norwegian industrial metal band "Red Harvest".
Here's the spectrogram of the FLAC file:

I use foobar2000 as a front-end with the latest LAME version (3.99.5), downloaded from RareWares. In the "Converter Setup" I choose MP3 (LAME) at 320 kbps CBR as output file format. After the encoding finishes what I get is an MP3 file with a spectrogram that surprisingly looks like this:

Why is the frequency cutoff at 14-15 kHz, this looks worse then most 192 kbps encodes?! If someone checked this he would probably think this is a re-encode or web-rip. Now I guess some people might say this is theoretically possible, depending on the specifics of the music the encoder might decide to filter those frequences and so on, but this is clearly not the case and here's the proof (surprise, surprise). When I use the same encoder and settings, but a different front-end (All2LAME) I get an MP3 file with this spectrogram:

So is foobar2000 doing something wrong or am I missing something or what? blink.gif
Same thing happens with some other songs by Red Harvest, but not all - for most of them it's the usual for 320 CBR ~20 kHz cutoff, but for some like this one it strangely drops to almost 14.

This post has been edited by ChesterB: Feb 10 2013, 03:54
Go to the top of the page
+Quote Post
ChesterB
post Feb 10 2013, 04:00
Post #2





Group: Members
Posts: 35
Joined: 2-February 07
Member No.: 40241



Ha, guess what, I decided to convert the MP3 file produced by foobar2000 to WAV and check its spectrogram and here it is:

So after all it is OK and it's Acoustica that's showing me a wrong spectrogram for the MP3 file for some reason, isn't it? If so - why, and why doesn't that happen with the MP3 file produced when using All2LAME as a front-end?

This post has been edited by ChesterB: Feb 10 2013, 04:01
Go to the top of the page
+Quote Post
Kohlrabi
post Feb 10 2013, 08:47
Post #3





Group: Super Moderator
Posts: 1114
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (ChesterB @ Feb 10 2013, 03:53) *
Why is the frequency cutoff at 14-15 kHz, this looks worse then most 192 kbps encodes?!
It looks like the psychoacoustic model decided there was nothing of value beyond 15kHz, and the encoder acted on it.

QUOTE (ChesterB @ Feb 10 2013, 03:53) *
If someone checked this he would probably think this is a re-encode or web-rip.
I really don't care if your files look good enough in a spectrogram to satisfy fools on piracy sites.

QUOTE (ChesterB @ Feb 10 2013, 03:53) *
Now I guess some people might say this is theoretically possible, depending on the specifics of the music the encoder might decide to filter those frequences and so on, but this is clearly not the case and here's the proof (surprise, surprise). When I use the same encoder and settings, but a different front-end (All2LAME) I get an MP3 file with this spectrogram:
LAME has an encoder switch to disable the lowpass, it's probably enabled in foobar2000 and disabled in All2LAME.

QUOTE (ChesterB @ Feb 10 2013, 03:53) *
So is foobar2000 doing something wrong or am I missing something or what? blink.gif
You're missing one important point: Spectrograms don't tell you whether the choices the lossy encoder made are audible. Is the lowpass audible to you? You can try to ABX the foobar2000 encodes with the lossless versions.

QUOTE (ChesterB @ Feb 10 2013, 03:53) *
Same thing happens with some other songs by Red Harvest, but not all - for most of them it's the usual for 320 CBR ~20 kHz cutoff, but for some like this one it strangely drops to almost 14.
See above.

This post has been edited by Kohlrabi: Feb 10 2013, 08:50


--------------------
It's only audiophile if it's inconvenient.
Go to the top of the page
+Quote Post
ChesterB
post Feb 10 2013, 12:41
Post #4





Group: Members
Posts: 35
Joined: 2-February 07
Member No.: 40241



I expected an answer like this. Look, I actually agree with what you say, I'm not arguing about that. It's just that I don't think this is the reason in THIS particular case.
QUOTE
I really don't care if your files look good enough in a spectrogram to satisfy fools on piracy sites.

What I said about the web-rip was just used as a comparison, piracy has nothing to do with it. Noone will be examining how those spectrograms look. And I'm not saying whether I will be able to hear any difference or not, this post is not about what is audible to me. I'm just trying to find out why there's this difference in the spectrogram of an MP3 file produced by the same encoder with the same settings but a different front-end. I'm not using any extra switches, so I don't think it's this either. All that I type in All2LAME is "-b 320" so I think that all it's supposed to do is pass that to the command line application lame.exe, it shouldn't be enabling any switches on it's own, neither should foobar2000, at least not without this being indicated somewhere so that the user knows.
Anyway, I isolated the problem. I opened the foobar2000 MP3 with a different sound editor and the frequences reach 20 kHz, so it must be Acoustica that's displaying it wrong:

Still, thanks for your reply.

This post has been edited by ChesterB: Feb 10 2013, 12:45
Go to the top of the page
+Quote Post
[JAZ]
post Feb 10 2013, 16:44
Post #5





Group: Members
Posts: 1797
Joined: 24-June 02
From: Catalunya(Spain)
Member No.: 2383



@ CesterB: If you look again at graphs 2 and 3 (especially the channel on the bottom), it seems clear that it is showing the same thing, but with a different range.

I don't know Acoustica, and I don't know if it lets zoom in and out in the frequency range (Audacity allows this. Also, Audacity allows to define the default range and depth of the graph).

I can't elaborate on why it happens on the foobar encoded one (if it happens always) and not the other one, but you can clearly conclude that it is a display error and not an encoding error.


@Kohlrabi: One thing is avoid to get wrong information from using the wrong method (audible problems using a spectogram), and another is to not even look at them. (And making assertions that, in this case, are disproved by the graph).

Also, I think you jumped a bit too early on the "don't pirate" bandwagon. "One would say this is a web-rip" is not the same than "I'll post the rip on the web".

Other than that, the rest of your post is correct.
Go to the top of the page
+Quote Post
Porcus
post Feb 10 2013, 16:59
Post #6





Group: Members
Posts: 1959
Joined: 30-November 06
Member No.: 38207



QUOTE (Kohlrabi @ Feb 10 2013, 08:47) *
You're missing one important point: Spectrograms don't tell you whether the choices the lossy encoder made are audible.


While this is obviously a point in its own right, it is a secondary issue when it comes to the question at hand, which is whether something is wrong in the process, given that two files generated by two applications which supposedly are front-ends to the same software, look different.

From the OP's question, it would be perfectly legitimate if (s)he even asked “why aren't the audio parts bit-identical?”


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
Porcus
post Feb 10 2013, 17:04
Post #7





Group: Members
Posts: 1959
Joined: 30-November 06
Member No.: 38207



ChesterB, a few tips if you haven't checked out already: “Godtech” and “Wounds”.


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
ChesterB
post Feb 10 2013, 17:21
Post #8





Group: Members
Posts: 35
Joined: 2-February 07
Member No.: 40241



Those are fine, is there a reason you picked them exactly? Anyway, at least I know it's a display issue as [JAZ] said, not an encoding one. Thank you, people.

This post has been edited by ChesterB: Feb 10 2013, 17:22
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 27th November 2014 - 06:52